Prognostic profile of systemic sclerosis: analysis of the clinical EUSTAR cohort in China

Patients clinically diagnosed with SSc between Feb 2009 and Dec 2015 were prospectively recruited from the European League Against Rheumatism (EULAR) Scleroderma Trials and Research Group (EUSTAR) and Chinese Rheumatism Data Center (CRDC) database of the Peking Union Medical College Hospital (PUMCH). Diagnosis of SSc was fulfilled according to the 2013 American College of Rheumatology (ACR)/EULAR criteria. Ethics committee approval was obtained for the EUSTAR and CRDC study and all subjects provided informed written consent.

Patients were mainly followed up by outpatient visits at intervals of 6 months to 1 year. Demographic, clinical and laboratory data were collected and entered into a database according to consolidated regulations. Patients whose last follow-up date was more than 1 year from May 2016 were followed up by telephone.

Patients were classified into limited and diffuse cutaneous subsets based on the definition of Leroy et al. [12]. Identification of peripheral vascular involvement included Raynaud’s phenomenon (RP), fingertip ulcer, loss of finger pads/pitting scars and telangiectasia. Identification of lung involvement included interstitial lung disease (ILD) and pulmonary arterial hypertension (PAH). ILD was defined as ground glass opacification or fibrosis on high-resolution computed tomography (HRCT). PAH was defined as a mean pulmonary arterial pressure > 25 mmHg at rest, together with pulmonary capillary wedge pressure < 15 mmHg determined by right heart catheterization or pulmonary artery systolic pressure (PASP) > 40 mmHg at rest based on an echocardiogram test. Those who only took echocardiogram test and hadn't taken right heart catheterization once in series and had maximum tricuspid regurgitant velocity (TRV) and pulmonary artery systolic pressure (PASP) of 2.9–3.4 m/s and 37–50 mmHg, respectively, were ruled out as diagnosis of PAH could not be confirmed [13]. Cardiac involvement included arrhythmia, left ventricular dysfunction (LVEF< 50%), decline of left ventricular diastolic function, pericardial effusion and valvular disease that could not otherwise be explained. Gastrointestinal involvement, according to the definition in the EUSTAR database, including esophageal (based on the presence of heartburn, regurgitation or dysphagia symptoms), gastric (based on clinical symptoms of early satiety, flatulence and vomiting) and intestinal involvement (diarrhea or constipation together with pseudo-obstruction secondary to small bowel involvement), in the absence of other explainable causes like treatment, cancer, etc.

Age at disease onset was defined as the age at first non-RP SSc manifestation, and age at disease diagnosis was defined as the age when SSc was diagnosed by a specialist. Length of follow up was defined as the time between the date of SSc diagnosis and last visit or death. Disease duration was defined as the time between the first non-RP SSc manifestation and last visit or death. Patients who were ≥ 60 years old at the time of disease onset were defined as having late-onset SSc.

Survival status was determined through the end of May 2016 based on database records or telephone tracing of patients for whom no data in the database had been entered for ≥ 12 months. The final status of loss to follow up was defined as having no data entered for ≥ 12 months with a failure to contact the patient on least two attempts. Data on cause of death were collected via medical records for patients who died in the PUMCH or for whom death certification was provided by relatives of patients who died in other hospitals. Deaths attributable to SSc were based on evaluation by three experienced rheumatologists from PUMCH.

The data were analyzed using SPSS 19.0 and differences between groups were analyzed by analysis of variance (ANOVA), the Mann-Whitney test or the chi-square (χ²) test, depending on the distribution of the variables. Bivariate odds ratios with 95% confidence intervals (CI) were calculated. Kaplan-Meier analysis was used to estimate survival from the date of diagnosis, with the Mantel-Haenszel statistic (log-rank test) used to analyze differences in survival. Cox analysis and logistic regression analysis were used to obtain independent risk factors for SSc prognosis.

A total of 448 patients with SSc were recruited for this study and 90.4% (405) were female. Limited cutaneous SSc (lcSSc) (254, 56.7%) was more common than diffused cutaneous SSc (dcSSc) (194, 43.3%). Overlap syndrome was present in 36 patients (8%). The mean age at disease onset and diagnosis was 39.0 ± 12.5 (8–75 years) and 42.8 ± 12.1 years, respectively. A significantly greater number of patients experienced disease onset at age< 60 years (427 vs. 21). Across 2167 patient-years, the median disease duration and length of follow up was 7.0 years (0.01, 49.4) and 53.5 months (0.25, 365), respectively. The majority of the cohort was Han Chinese, but 11 ethnic minorities were also included (4 Manchus, 5 Mongolian and 2 Korean). Past history of cardiovascular and respiratory diseases was as follows: 47 patients had a history of hypertension, 9 had previous coronary heart disease, 6 had previous cerebrovascular disease, 1 had rheumatic heart disease and 1 had silicosis.

Raynaud’s phenomenon was the most common peripheral vascular involvement in patients with SSc in this cohort (424, 94.6%), 72.3% of whom (315) presented with Raynaud’s phenomenon as the first disease manifestation at disease onset. More than half of the patients (271 patients, 60.5%) had different respiratory symptoms, with shortness of breath after exercise (250, 55.8%) being the most common, followed by cough (106, 23.7%) and dyspnea (38, 8.5%). Most (358 patients, 80.1%) were diagnosed as having SSc-ILD at baseline, whereas 55 had concurrent ILD and PAH.

Pulmonary function test (PFT) had been undertaken in 80.8% of patients (n = 362) at baseline, and the mean ± SD for forced expiratory volume in 1 s (FEV1), total lung capacity (TLC), forced vital capacity (FVC), diffusing capacity for carbon monoxide (DLCO)% on PFT were 82.1 ± 15.4, 7.8 ± 17.5, 81.2 ± 17.0, and 62.2 ± 20.0, respectively. Other organs involved at baseline are shown in Table 1. Elevated erythrocyte sedimentation rate (ESR) occurred in 121 patients (30.9%). C3 and C4 levels at baseline were 1.03 ± 0.25 g/L (in 314 cases) and 0.20 ± 0.09 g/L (in 312 cases), respectively. The average immunoglobulin G (IgG) level was 15 mg/dl (6.5, 52.2). Antinuclear antibodies (ANA) were seen in 97.7% of patients (336/344), whereas the presence of anti-ribonucleoprotein (RNP) antibody was less frequent (90/344, 26.2%). Anti Scl-70 and anticentromere antibody (ACA) were present in 46.8% (169/361) and 16.5% (48/291), respectively, of patients, 3 of whom were double positive for anti Scl-70 and ACA.

During follow up (53 months (0.25, 365)), 24 patients developed pulmonary interstitial diseases, and the prevalence of ILD reached 85.5%. PAH was identified in 78/361 patients, accounting for 21.6% of the cohort, and 63 patients had both PAH and ILD. Arrhythmia was present in 29 patients: among those with arrhythmia, only 5 patients experienced palpitations and other symptoms. At baseline, one patient had ovarian cancer and one had thyroid carcinoma, whereas two patients developed tumors during the follow-up period, including one with lung adenocarcinoma and one with leukemia.

Among 402 patients, 348 of them took glucocorticoid during the disease course: 374 patients received immunosuppressors and 43 of them took only immunosuppressors; 11 patients took neither glucocorticoid nor immunosuppressors. Among patients who used immunomodulators during follow up, 250 patients took cyclophosphamide (CTX), 2 took cyclosporin (CsA), 28 took mycophenolate mofetil (MMF), 5 took tacrolimus, 2 took azathioprine (AZA), 9 took leflunomide (LEF), 103 took TwHF, 108 took MTX and 29 took hydroxychloroquine (HCQ); 102 patients used two immunomodulators at the same time and 19 used three immunomodulators at the same time.

The 448 patients with SSc were followed up for a total of 2167 patient-years. The median survival time for the patient cohort was 53 months (0.25, 365). Survival rates from disease diagnosis were 97.0%, 94.6%, 91.1% and 87.8% at 1, 3, 5 and 10 years, respectively (Fig. 1). The cumulative survival rates at 1, 3, 5 and 10 years in patients with lcSSc were 97.6%, 94.2%, 91.1% and 89.1%, respectively, whereas for patients with dcSSc the rates were 96.3%, 95.1%, 91.2% and 87.0%, respectively.

No significant differences were seen in subgroup survival analysis according to subtypes (log-rank test 0.002, p = 0.968), baseline vascular (log-rank test 2.407, p = 0.121), joint (log-rank test 2.868, p = 0.090), muscular (log-rank test 1.009, p = 0.315), gastrointestinal (GI) involvement (log-rank test 0.027, p = 0.870), gender (log-rank test 0.014, p = 0.907) and age of disease onset (log-rank test 0.249, p = 0.618) (Additional file 1: Figure S2).

The median modified Rodnan skin score (mRSS) was 6 points (0, 43) in this cohort. According to the distribution of skin scores, 87.5% of patients had a skin score < 16 points. Patients with higher skin score (mRSS > 15) had a poorer prognosis (log-rank test 7.977, p = 0.005). However, survival was better in patients who had no respiratory symptoms (Additional file 1: Figure S1) and less pulmonary dysfunction (FVC% < 70%, log-rank test 14.58, p = 0.000, DLCO% < 60%, log-rank test 23.58, p = 0.000) (Fig. 2).

No significant differences were seen in subgroup survival analysis according to use of HCQ (log-rank test 1.186, p = 0.276) or MMF (log-rank test 0.016, p = 0.9). TwHF (log-rank test 7.851, p = 0.005) and MTX (log-rank test 7.925, p = 0.005) were found to be protective against mortality in survival analysis. Patients who took CTX during the disease course had a poorer prognosis (log-rank test 5.177, p = 0.023).

Patients who had PAH (log-rank test 32.96, p = 0.000), arrhythmia (log-rank test 14.28, p = 0.000) and pericardial effusion (log-rank test 6.879, p = 0.009) at baseline had a poorer prognosis than those who did not. However, there was no significant difference between patients with or without ILD at baseline (log-rank test 0.303, p = 0.582). Among 288 patients who had both HRCT-confirmed ILD and PFT simultaneously, those who had pulmonary dysfunction (defined as FVC < 70% or DLCO < 60%) had a poorer prognosis (log-rank test 16.266, p = 0.000) (Fig. 2).

Among 358 patients who had ILD at baseline, 217 of them had continuous monitoring by HRCT. We qualitatively assess disease progression according to the description in the radiology reports: 64 patients had radiographic evidence of aggravation (median disease duration at baseline was 0 (0, 132) months) during the disease course, and 153 patients remained stable or improved (median disease duration at baseline was 1 (0, 165) month). A poor prognosis was not confirmed among patients with ILD who had radiographic aggravation (log-rank test 2.629, p = 0.105). There were 89 patients who did not have ILD at baseline, and 24 of them developed ILD during the disease course. New onset of ILD was not associated with a poor prognosis (log-rank test 0.111, p = 0.740).

Univariate survival analysis suggested that SRC, PAH, arrhythmia, pericardial effusion, skin score > 15, FVC% < 70% and DLCO% < 60% on pulmonary function tests (PFT), and respiratory symptoms at baseline were associated with poor prognosis. Further Cox multivariate analysis confirmed that PAH (hazard ratio (HR) 6.248, 95% CI 2.855, 13.674) and arrhythmia (HR 4.729, 95% CI 1.588, 14.082) at baseline are independent risk factors for poor outcome in patients with SSc.

Among the cohort, 40 patients with SSc died (Table 2): 27 deaths (67.5%) were related to SSc. PAH was the leading cause of death, accounting for 55% of all deaths. The mortality rate in patients who had PAH during the disease course was 28.2% (22/78 patients). Of the 63 patients in this cohort who had both PAH and ILD, 17 died (mortality rate, 27.0%). Among the four patients who died from myocardial infarction and stroke, only one had no history of cardiovascular involvement or hypertension. The two patients who died of lung infection both had SSc-related lung involvement, one had PAH, and the other had ILD. One patient died of bacteremia due to cholecystitis. Two patients who died had tumors and one died of leukemia 5 years after the SSc diagnosis. The other patient was first treated for hemoptysis and within one year of SSc diagnosis this patient died from lung adenocarcinoma that had metastasized to the bone, meninges and liver.